ClawHub
Back to skill

Security audit

PowerShell Advanced

Security checks for vulnerabilities and agentic risk

Overview

This PowerShell reference fits its stated purpose, though its process-handling examples have reliability limitations.

Treat this as a coding reference, not a safety guarantee. Correct the sequential process-output reads and bound timeout cleanup before production use. Send tokens only to intended HTTPS endpoints, and review permission changes before running examples.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

The process-capture example can deadlock despite being documented as capture without deadlocks.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

Reading stdout to completion before reading stderr can deadlock: if the child fills the stderr pipe while stdout remains open, the child blocks writing stderr and the parent blocks waiting for stdout EOF. Although the specific git example may not trigger this, presenting it as a deadlock-free general template encourages reuse that can cause availability failures, especially with attacker-influenced child output; the example also lacks a timeout.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.